Transcription
In this video, I will share with you a tiny bit of what I learned during this week. Well, one thing in particular that I learned during this week. So, you can see this video a bit like a documentation. Okay, of my learning. By the way, I tell myself that it would have been truly incredible if, since the beginning of my channel, every week, I had made a video to document my learning. It would have been incredible. There, you would have been able to see, in fact, you would have known exactly everything I learned since the beginning and in what order. In short, it would have been incredible but it would have also been good for me to have a record of my journey, but well, it's okay.
So now, I'm going to talk to you about a concept that I discovered, well, I truly discovered it this week. Uh, no spoilers, no holding back, it's the op, it's the op. And in fact, I'm going to explain the context a little. At the beginning of the week, I stumbled upon this video and I watched it, you see. So this video claims that we should stop using N8N because there is a tool that is 10 times better, 10 times faster, to not waste your time. It's a tool that allows you to generate workflows. Well, not on N8N, it's a separate tool. And basically, you talk in the chat a bit like the no-code tools, and it allows you to generate automations. Okay? It generates a kind of, it asks you questions and it generates a kind of workflow on the right. I don't know if you can see it well here, but there you go, a kind of workflow like this.
So, my first impression when I saw this video was that no, this tool will not replace N8N and will never surpass N8N. First of all, for the first reason, which is that it's all well and good for simple automations, but the technical complexity that we can get into using N8N. But this tool will never be able to do that. That's one. Secondly, this tool does not run locally. That's what you're going to see, that's what some comments will say. N8N, we can deploy it locally, it's still much safer. So here there's security, there's also cost, the financial aspect. And if we go down a little, I'll see this comment that says, uh, great but two remarks, N8N will do it for sure. Uh, that's true, by the way. N8N will necessarily do it sooner or later. Two, you use the word "agenda" extremely abusively. Well, that's everyone. Three, it's not open source. That's what I'd like to focus on. It's not open source.
So, what is open source? Uh, I asked ChatGPT. I asked ChatGPT to give me a definition of open source. So, there you go, ChatGPT tells me that open source is a development model in which the source code of the software is accessible to everyone. This means that anyone can consult, modify, improve, or adapt the software to their needs. For N8N, this means that the community can directly contribute to the tool, create its own modules, and that it is not dependent on a company for every evolution or correction. So, the fact that we can host N8N locally, on our own computer or on a server, that's already a characteristic of open source. Okay? Because in fact, if it's possible, it's precisely because N8N's code is publicly available. Okay? So we can directly take N8N's code and run it on our computer. So N8N will run directly on our premises. So N8N will take care, well, the N8N company will take care of very little regarding the power. Okay, the power we need to run our workflows, the security, the updates, etc. For example, updates, we will have to do them manually ourselves, given that we are hosting N8N ourselves. Whereas if we used the paid cloud version of N8N, well, N8N would have taken care of everything. The workflows, etc., would not have run thanks to our computer, the power of our computer, but rather thanks to the power that N8N makes available, the power of their machines, their servers. They manage everything related to security. Well, you still have to be careful, you have to take measures, but they manage it. But also updates, well, they take care of that. You see, all we have to do is maybe click on a button to update, and even some will be done automatically. So there you go, all of this is a characteristic of open source.
So, open source, which is the opposite of closed source. I can call it that. I don't know if it's a term that exists, but I think I've already seen it. So open, closed. Okay, it's the opposite. A closed-source tool, for example, will be all the SaaS you can see on the internet, which we access by paying. By the way, Make. Okay, stay within the context. Make is a source tool, if I can call it that. Why? Because in fact, we cannot use Make locally. We are forced to pay for Make, you see. There is no option to self-host Make. The software we are talking about in this video. By the way, I'll see if the link is in the description. So, there you go, the link is there. Well, this software is also closed source, in fact, because we don't have the opportunity to self-host it. The only way for us to use it is to pay or to stay within the limits of the free plan. And in fact, as specified by ChatGPT, the open-source nature allows for community contribution. Okay? Since the software, the software's code is publicly available, well, all developers can simply contribute to the project. There you go, they see the code of the application, how the application works, and they can simply say, "Here, I want to add this or that feature." They contribute to the project. So contribution is an essential characteristic of open source.
Uh, by the way, the source code of N8N, you can find it. If you type N8N, normally it's on GitHub. So there, there you go, this is all the code of N8N. Okay. All this is the code of N8N. When you install N8N on Docker, well, it's all this code that you install. Uh, I have, for example, Hostinger, Hostinger N8N, a Hostinger VPS, N8N. By the way, many people think that this is a cloud version of N8N. It's not a cloud version of N8N. It's a version, let's say, self-hosted, but just not local. It's a self-hosted version, but it's remote. How to explain it? In fact, we can say that there are three versions of N8N. Let's say there's the cloud version. Okay, it's the version on the N8N website, you pay €20 per month, I think. That's the cloud version. Then, the second way to use N8N is self-hosted locally. So you host it yourself locally. And that's, for example, on Docker. So you have Docker on your PC. Uh, I made a tutorial for hosting N8N on Docker, but basically, you host it on your own computer through Docker. You see, so you don't need to pay, as I explained earlier, all the advantages I explained earlier.
And the third way, let's say, to use N8N is, well, self-hosted but on a remote server. Okay. So basically, what you do, and what most people do, and perhaps they don't know it, is that with Hostinger, which you see here, for example, Hostinger is one provider among many others, you subscribe to what is called a VPS. Okay? What's here is a VPS. VPS means Virtual Private Server. It's a virtual private server that you have access to virtually, even though the machines are physical somewhere in the world, and in fact, they rent it to you. Hostinger will rent you this machine. It's a computer that is somewhere in the world. Hostinger provides it to you, and this computer has these specifications. So it has two CPU cores, 8 GB of RAM, 100 GB of disk space, bandwidth is this. So in fact, they provide you with a computer that is just distant, but you can benefit from the power of this computer to run your workflows, to run whatever you want. By the way, I specify, and we'll talk about it later in the video, not just N8N workflows, whatever you want. In fact, it's just a computer. Okay? A private machine that they rent to you, that they lease to you, because you still have to pay, but you have the opportunity on this machine to install, for example, N8N, but that's what you do, for example, when you subscribe to this, and that's why Hostinger is quite popular for this, because in fact, they offer us the opportunity to directly install N8N on the machine. Okay? So basically, you subscribe to the machine, the computer, and Hostinger, if you select N8N during registration, well, they install N8N directly for you. They were able to set this up by default because they have access to N8N's code. So they created a kind of template to make it easier for people, so that there isn't too much technical complexity, you see.
And so, perhaps you are asking yourself, or you have already asked yourself, why use a VPS? Knowing that I can use, for example, Docker, you see, I can use Docker Desktop which runs directly on my PC, and there I wouldn't even have to pay for a virtual machine because it's my PC, you see. Well, the reason is simple, well, one of the reasons is that if I run it on Docker, I am forced to leave my PC on. I am forced to, because, well, Docker, it's the containers that run in it, and the containers need to be running for the workflows to work. So, in fact, if you have no need to run your workflows actively, that is to say 24/7, you see, to keep them active, your workflows, even when you sleep, well, you can just use Docker. You start your Docker, you start your Docker containers when you use N8N. By the way, if you don't know what a container is, I talk about it in the video for installing N8N via Docker, but basically, a container is just a virtual space where we run N8N. In short, I think I've explained that quite a bit there with a good analogy that I found thanks to ChatGPT. You should watch the video. Well, I'll try to put it in the description. But there you go, the disadvantage of running, of using Docker, is in fact that your computer must be on all the time. If you want your workflows to run all the time, well, your computer must be on 24/7, which is not possible.
So, it is much more sensible. If you need workflows that run all the time, well, opt for something like Hostinger. So on a remote server, these remote servers, well, they will be on all the time. You see, they function like that. It's by default, it's like that. So you can turn them off in the settings, but by default, well, they work all the time. So if you need, well, a workflow that runs every minute, that checks your email every minute, it's much more relevant, in fact, to opt for that, you see, to opt for a private server. In short, this is a bit for context. Now, let's talk about the truly open-source aspect, which is the discovery I made this week, by the way. So here, I'm going to go into my N8N, which is actually on Hostinger. Why did I choose Hostinger? Because, in fact, to avoid technical complexity, as I told you, they offer a template. I didn't really have any experience in managing a virtual machine, etc., etc. So to start smoothly, I told myself, I'll just go for Hostinger, you see, and not try to be clever, you see, not try to be an expert, just like that.
So here, I'm putting things in context. This is a project I'm doing for companies. Well, I won't explain what the project is, but check this out. Here, I realized that I had a problem. I needed to analyze an image to know what was on it. So we have an administrative document, and I need to know what administrative document was sent to me by email. Okay, it could be, for example, an ID card, it could be, I don't know, a driver's license, whatever. So this is sent to me by email. You see, thanks to this Gmail, I check the email I received, whether in fact, it's this document, and in fact, I give the AI the ability to analyze the image and tell me, in fact, what it is, what's on the image. Is it a driver's license? Is it an ID card, etc. So, I'll put that aside for now. I'll run this, and there you go, just to make it more fun. Okay, you see that there are two files here. There are two files. Uh, and then there you go. So what I did here, well, before putting the AI node, what I had done initially was to use the Open AI node. Okay. So here, by the way, to explain why I structured it like this. So here, basically, in the first node, I receive the email with the files. Okay? You see the binary files here. Then I'll use a filter to make sure, in fact, to only let through emails with binary files. So here it's a single email, but let's say there were three emails retrieved, I'll only let through those that contain binary files because I don't care about the rest. I just want those that have binary files, meaning files so that I can analyze, well, what file is inside.
Then, I used a code that will allow me to extract the name of each file. Okay? Because the name, it was returned to me here. I don't have the ability to drag and drop it into a field. Okay? So, I asked the AI to generate this code for me, explaining the situation. By the way, I'll make a video explaining how I use code because I never code myself, but how I think, when I think of using it, and how I use it with AI. In short, here in the AI Basic LM chain node, I will simply ask it to determine if this email is a file submission from a client for their administrative file and to respond only with yes or no, okay? And there, for it to be able to do the work, I give it the subject of the email, the body of the email, as well as the files, the names of the attached files that were generated by the previous code node. Okay? So it will have to, basically, analyze the subject of the email, the body of the email, and also the file names and estimate, in fact, whether it is a file submission. Because I imagine that if someone submits a file, they will say in the email subject, "File submission," or in the email body, they will say, "Please find attached my ID card, for example." In short, I give the AI all this information, as well as the file names. For example, if it's an ID card, maybe it will be "identity Ousman," something like that. So there you go, I give it all the information so that it can know if it is indeed a file submission or not. And I tell it to respond with yes or no. Okay. So there, it responds with yes or no. And then I will go into a branch depending on whether it's yes or no.
So here, I will use the code node again. Why? Because I want to retrieve the binary data, you see? I want to retrieve the file to be able to give it to the AI and for it to analyze what administrative file, administrative document it is. So, I use the code. Why? Because in fact, I cannot directly give the AI a file. I will link this to everything related to binary files. I can only use it if the binary file is in the previous node. So here, you can see, for example, that in the Gmail node, there are binary files that were returned to me. I see them here in the N8N, the next node, I can see them here. Okay? I can also use them. But if I go into a later node, look, if I go into the Gmail node, I don't have access to it. Okay? Or if I go into the code node here, you can see that if I go into filter, I don't see the binary file. If I go into the Gmail trigger, I don't see it. By the way, to see it, it needs to return a binary field here, as is the case here, because basically, what this code node allowed to do is to load the binary files because, once again, to use a binary file, it must necessarily, this is a particularity of N8N, but it must necessarily be in the node that is just before. So I had to retrieve what was returned to me in this node as a binary file, in the Gmail node, and load it, basically, load it to have it returned by the code node here, so that I can then use it in the AI agent and tell it to analyze it. Uh, and then there you go, again, I didn't generate this code, but well.
So, this returned the data in two items because there are two binary files. And in the first instance, what I tried to do is, in fact, to use the intuitive Open AI node that allows you to do that. By the way, it's called "analyze image." What does "analyze image" do? Well, it analyzes an image, simply. Except that for me, the problem I encountered, first of all, was that I didn't have an image. What I have is a PDF file, so a priori it shouldn't work. But I tried anyway with an image file and I found that this node didn't work. I don't know why. I'll try to do it again. So I'll put, for example, an N8N form here, a form event file, I'll put the image. Okay, the element type, I'll put file, and here I'll use an Open AI analyze image. It says "What's in this image?" Uh, GPT-4 or Mini, then binary file, because the input type is not a link, we have direct access to the image with the binary file and not through a link. The input data field name, how did I name the file? I named it "image." So a priori, here it's "image" that I should put, and then there you go.
So here, I'll execute this. Uh, here, I'll provide it with an image. Submit. You can see "The item has no binary field 'image'." Okay, so in fact, I have to write it like this, exactly. So with an underscore, they replaced it with an underscore. I execute this. You can see there's an error. So, I did some research on this error. I didn't find what it meant. I couldn't debug it, in fact, simply. So I told myself, in any case, this node doesn't work. Okay? This node doesn't work, whether with a PDF or with an image. So in any case, it's an image, uh, there you go, I can't use it. So what I told myself, I found an alternative. I told myself, I'll use another LLM that has vision capabilities. Which LLM has vision capabilities? Open AI, it can do it. Okay, because when you go on ChatGPT, you give an image in the chat, you can discuss the image with ChatGPT. To my knowledge, Gemini can also do it. So I told myself, I'll try to go through Gemini, and that's what I did. Now, Gemini, you don't have an "analyze image" action like ChatGPT does. If you type Gemini, you see, you'll see that there's only "Google Gemini Model." So that's the small model that you can only give to an AI agent. So I have no choice, I'll use an agent, but it's not so bad. Why? Because there's this option "Automatically pass through binary images." So basically, we'll also give the binary file as input to our agent. Okay? So this option, we come here, we click on "Add option," we click on this, and we activate it. "Automatically pass through binary image." So if there's binary file data in the previous node, which is the case, it will give it as input, you see.
So here, I'll re-execute. You'll see what it should give normally. So there you go, here it returns two items. So each one, well, a description for each item. Now, here we can see what it gave me. "Please provide with the file." Okay. So here, basically, it's as if I hadn't given it anything. You see, why does it seem like I haven't given it anything? Because it only passes images, but not files. I'll confirm this theory, by the way. I'm not sure, but I think so. I think that's why it couldn't analyze it. Uh, I'll put a form again where I'll put a file, an image file. Now, I'm sure that's the case because yesterday, I had tested, uh, the image. Click, click. Now, I'm sure that's the case because yesterday, I had tested and it was the case. So here, I'll give it this image. Basically, it's a screenshot of a book. Here, normally, the prompt is "What's in this file?" So, what's in this file? It tells me, "This is a portion of a book, specifically chapter 6 of the book Essentialism," and it explains what's on these two pages. In short, there you go. So it was able to analyze what's inside.
Well, the problem was actually that it wasn't an image, and that this option, in fact, it had to be an image. From there, I think another problem occurred. In fact, the problem that occurred is that this workflow will be used by a person who will receive PDF documents mostly. This means that the option I have to allow Gemini to analyze the file, it absolutely must be an image in its entirety. However, I have PDFs, I have PDFs that are provided to me. So I told myself, necessarily, in the workflow, it will be necessary to convert these PDFs into images. It's possible. There are many, there are many tools that allow you to do that. Okay, there are many tools that allow you to do that. You can see, for example, this one, I've used it for years. It allows you to convert files. All you have to do is click here, choose the files. You give your file. So, I'll give you this one, for example. Here you say in which format you want to convert it. PDF, for example, you click on convert, and it will convert it. Now, this is a manual method. I am in the context of automation, which means that I would need an API. Conversio, which you see here, has an API. API. You can see that Conversio has an API, which must be paid. I think it's paid. Most of these APIs are paid.
So, basically, what I did is that I went on Google, I typed "PDF to PNG API," you can see, I did the search. Look, I searched all these sites, and looked for an API that allows you to do that because I want to do it through automation. But well, all the APIs I came across are simply paid. And it's actually there that I discovered the power of open source. It's there that I discovered the power of open source. In reality, what I observed is that all the software you can see on the internet, all the SaaS you see there, they all use the same open-source software. As I was telling you, open-source software is software that is free, that you can install on your machine and do whatever you want with. So, basically, what do these people do? They use APIs, if they haven't developed it themselves, but they use APIs that are open source, and basically, they've faced a bit of technical complexity and they've created a SaaS, and so we, non-technical users, well, they sell us these SaaS, you see, whereas behind it, they use something that is free, that is open source.
So, in fact, I'm having a realization, but if we are capable of using open-source tools ourselves, well, it's extraordinary, it means we have access to all software, to the source, because almost all software exists for all uses. So, basically, we just need to be able to use open-source software. And in fact, I had the realization when I discussed a bit with ChatGPT and explained my problem to it, and it told me about this software, PDF to PPM. PDF to PPM. It's software that we use on the command line. The command line, it's this. It's this command line. Uh, there you go, it's this command line. It's software that we use on the command line and that allows you to do this, in fact, that allows you to do the conversion. It's simply thanks to a simple command, by the way, that ChatGPT gave me. But thanks to a simple command, you can do this file conversion directly. Okay? And it will be free, it's open source. That is to say, the software, you'll just have to install it either on your machine, your computer, or on your Docker, or on my, in my case, on my Hostinger VPS. Okay? On my Hostinger VPS because, in fact, there are nodes on N8N that only work locally. You cannot use it if you have the cloud version of N8N, you can only use it if you have a local self-hosted version. Okay? I have self-hosted N8N on my Hostinger, or it can be Docker, there are many other software that allow you to do it.
So, in fact, you should know that there are software that only work locally, like these two. Okay, this one is not part of it, it's separate, but these two nodes, for example, they only work locally. "Execute command," it allows you to execute commands on your machine directly. So, basically, what I would have to do in the context of my project here is to take this "execute command" node, put it here, and directly, I will be able to put the command. The command, as I was telling you, the PDF PPM software that allows you to convert a PDF to a PNG. I will actually use a command in it that will allow me to convert the file. By the way, I'll show you the command that ChatGPT gave me. So there you go, it gave me this. You see, it's a command like this. So you'll put "PDF PPM PNG" something like that, in fact. You see, it would look like this. It would look like this. And what would it do, in fact? Simply convert my PDF to PNG, in fact, simply for free, locally, and I wouldn't have had to pay for a paid API on Google, etc., etc.
Do you realize the power, the power of open source? All the software you can see here, in fact, on the net, etc., well, they use open-source software to function, and then they charge us for the service. In fact, there's also a software called, I think I opened the page, I closed it, PEG. Basically, what it allows you to do is to process audio and video. It also works on the command line. By the way, all these most powerful software, they work on the command line. That is to say, there's no interface where you click. No, it's commands that you type. So, as I was telling you, either if it's on your PC, you can type the commands here. If it's not on your PC, well, if it's, for example, on N8N, you can simply type this in the "execute command" node. So it's extraordinary. For example, when I record my videos, they are in MKV format. I think YouTube doesn't accept MKV, or in any case, for one reason or another, I need to convert my files to MP4. And before, what I did was use expensive software, which is often paid, and then there you go, it's complicated. And when I discovered FFmpeg, in fact, I discovered that all these expensive software use FFmpeg in the background, you see, to allow us to do that. And then they pay for it, naturally. But well, if we are capable of using the software at the source, which is also free again, well, why not? And it will even run possibly faster, maybe. Well, it will depend on the power of your PC. It can potentially run faster than if you go through expensive software.
In short, you see the software that allows you to cut long videos, for example, long podcasts into shorts. Think about it, these software use open-source software like FFmpeg, in fact, you see? Because FFmpeg, in addition to video conversion, etc., it also allows you to cut and edit videos, to put images on them, to put text on them, etc., etc., to, once again, cut videos. So often they use this software, FFmpeg, to do everything, to cut long format videos into short formats, and then they sell that as a SaaS. So there you go, I just want to show you the power of using the local capabilities of N8N. But that's why, in fact, I realized that this will never happen. This will never happen with software that is not open source. N8N will never be surpassed by software that is not open source, for these reasons.
First of all, most non-technical people, they don't know this. What they see with the open-source aspect is the cost, the cost advantage, you don't pay compared to other software, but there are many other advantages, like what I'm showing you here. In fact, there are other nodes that are used only locally. So, you can see the node that I discovered and that I will certainly use in my project. It allows either to read files from my computer's disk or to write files to my computer. When I say read, it's basically retrieving the file. So, basically, I have a file on my PC. If I run N8N on my PC, I'm running N8N on my Hostinger. But if it's on my PC, I can, in fact, take a file from my computer, put the URL. So it's not, it's not
really the URL, the path we call it. So these two points / uh Ousman / download and so on and so forth. I can retrieve a file on my computer and do what I want with it in my automation. And then what I can do is write, so write the file again on my computer. So by placing a file on my computer, saving it on my on my machine on which N8N is running. So there you go, these are nodes that are used only locally and if they are less popular than the rest precisely because, well, certainly a bit, a bit because of the technical barrier, but there you go, there are many. You see, uh FTP for example here, it's a log, it's one that is used only locally. So, well, another thing I learned because, well, I told you that we are in documentation mode, you see. So I'm throwing out everything I've learned. Another thing that I realized is that the container is really isolated. So the container is what N8N runs in, it's really isolated from the rest. So, basically, what I had done is that I went onto my Hostinger and I went to my browser terminal. So the browser terminal is simply the command line of my hosting machine. The command line on my PC. That's it. Command prompt, it's simply my command line. It's my terminal. It's also called a terminal. And so with Hostinger, given that my VPS is remote, by the way, I don't know why it's not loading, but given that my VPS is remote, I don't have physical access to the computer, well, I have access to the terminal, that is to say, the command line through the browser. So there you go, this is the command line I have access to. Here, I can simply type commands. So, basically, what I did is I downloaded the software directly on the Hostinger command line because I told myself, my N8N is on Hostinger, so I'll go to Hostinger. By the way, just to clarify, these software programs that you can see here, they are only downloaded through the command line. Okay? Sometimes you have the opportunity, if you are on your own computer, you have the opportunity to install it like normal software by going to the website and clicking on download. But otherwise, there you go, I'm downloading it through the command line. So, to do something like ffmpeg for example, it would have been, I think it was something like sudo sudo apt install ffmpeg -y. Y, I think that was the command. And basically, it will install FFMPEG on my PC. I will be able to use it through the command line. If I want to convert a file, well, I can do it, create an automation that will allow me to convert my file directly. So, basically, what I can do is create my own SaaS. Okay? And my own API, even, that will allow me to convert my files, just like this application does. Except that I would have an automation for it. I would have my version of this software for myself, you see, and I would have created it all by myself and created it quite easily. But actually, what I realized is that it's really separate. Here, in fact, I am on my Hostinger VPS. Here, as I told you, I am on the terminal of my Hostinger machine. Okay, so the thing is that N8N, it runs on my Hostinger VPS, yes, but the problem is that N8N runs in a distinct environment. N8N runs in what is called a container, a Docker container. And so, a Docker container is separate from the rest, you see. So if I install FFMPEG directly on the terminal of my Hostinger VPS, what I actually tried to do is to subsequently use FFMPEG on N8N through the execute command node, the one in pink. Okay. I tried to use FFMPEG here, but actually I couldn't. Why couldn't I? Because, in fact, the N container, the container is isolated from the rest. So, in fact, what I finally had to do is to reinstall my, basically, basically, in simple terms, reinstall and update my N8N container by installing FFMpeg in it, by installing PDF or PPG, I think. In any case, all the software that I want to use through the command line on N8N, well, I was forced, in fact, totally forced to, in fact, update my container a little bit and reinstall it with the software. So, basically, the software must be available and accessible in the container because if you just put it on the VPS but it's not in the container, it's more difficult, in fact. So, well, I think that was all. I didn't even have a plan for this video, but there you go. I think I've covered it a bit, haven't I? So, there's the power, the power of open source. N10 will never remain first as long as N8N is open source and the competition is not open source. And well, there you go. So, well. So, well, this was a bit of a special thing, let's say, that I learned this week. I will continue to explore this, this, this open source character, this aspect, you see. And I will certainly, I don't know when, but make tutorials that will include, for example, FFM, FFMPEG. Well, in fact, I have an idea for a project that I want to do, and when I do it, well, I'll present it to you in a video, quite simply. But so, well, it will be a good opportunity to show a little bit how to use this, this command line, well, this execute command node on N8N, as well as other nodes like the read from file from disk node or something like that, the green node we saw earlier. By the way, earlier, I was thinking about something, I told myself, I'm going to use my Twitch channel, my Twitch channel to do streams where I'm just going to build, you see, build workflows live, you see. Even if no one is there, I'll start, and then, well, if someone wants to come and chat or wants to see my process, well, they can come, you see. So, if I don't forget, I'll put the link, the link to the Twitch channel in the description because I don't like doing YouTube lives. I prefer Twitch for that, and well, that's it, quite simply. Uh, well, don't hesitate to tell me what you thought of it. I hope the video wasn't too technical, but there you go, and that I've simplified it. But yeah. So, well, we'll say goodbye next time. Peace.